Sovereign AI for Families: One AI That Knows Your Whole Family

The question of AI in family life is no longer hypothetical. Millions of households already have at least one member using an AI companion daily — and the numbers are growing. The problem is that most of those companions know nothing about the family. They know the individual user. They have no idea that the user has a daughter who struggles with anxiety, a partner who works nights, or a grandmother who lives alone. Each companion is an island.

MEOK's Family plan was designed to solve exactly this. Not by collapsing individual privacy — everyone still has their own sovereign companion with their own encrypted memory that no one else can read. But by creating a shared layer on top: shared memory threads where families choose to connect, a Guardian safety network that watches for threats across the whole household, and a care architecture that treats the family as a living system, not a collection of isolated users.

This post covers everything you need to know: how the plan works, what Sovereign Memory means for families, how Guardian amber alerts function, what protections exist for children, and why the Maternal Covenant makes MEOK categorically different from every other AI product your family might be using.

What exactly does the MEOK Family plan include for £29 a month?

The Family plan is a single monthly subscription at £29 that covers up to five individual MEOK companions. Each companion is fully sovereign — its own memory, personality, conversation history, and privacy boundary. No two companions share data unless the users have explicitly consented to sharing a specific memory thread.

To put the value in perspective: five individual MEOK companions at the standard solo rate would cost significantly more. The Family plan compresses all of that into one subscription while adding capabilities — shared memory, Guardian amber alerts, family admin controls — that simply do not exist at the individual tier. The family-level features only become possible when the companions know they are part of a household.

What is Sovereign Memory and how does it work across a household?

Sovereign Memory is MEOK's core memory architecture. Every conversation, every preference, every piece of context that a companion learns about its user is stored in an encrypted memory graph that belongs entirely to that user. It is not stored on MEOK's servers in a form that MEOK can read. It is not used to train models. It does not follow the user to another platform. The memory is sovereign — meaning the user owns it completely, and the user decides what happens to it.

In the Family plan, Sovereign Memory gains a new dimension: shared threads. A shared thread is a segment of memory that two or more family members have agreed to make visible across their companions. An example: a family has an upcoming holiday to Portugal. Mum creates a shared thread for the trip — destination, dates, activity ideas, budget constraints. When Dad asks his companion about the holiday, it already knows the full context. When the teenager asks for Portuguese phrase suggestions, their companion knows the dates so it gives contextually accurate advice. No one had to brief each companion separately. The shared thread did the work.

The critical design principle here is that shared memory is additive, not reductive. It does not diminish individual sovereignty; it builds a family layer on top of it. The teenager's private conversations remain as private as they were before. The parent's work stress, health concerns, and personal reflections are not visible to their children. The shared thread is a deliberate, consensual window — not a surveillance channel.

How do Guardian family alerts work in amber mode?

Guardian is MEOK's safety layer. In solo mode, Guardian watches every inbound message for scam patterns, coercive control signals, grooming language, and crisis indicators. It scores every message from 0 to 100 and takes graduated action based on severity. In the Family plan, Guardian operates in amber mode — an enhanced configuration that adds a family-level alert network on top of the existing individual protections.

Amber mode means that when any companion in the family group detects a HIGH or CRITICAL threat, designated family members receive a silent notification on their own companion. The notification does not expose the content of the threat — it simply tells the designated guardian that a protective response may be needed for a specific family member. The guardian can then make a real-world decision: call them, check in, or contact emergency services if warranted.

The Guardian companion in the Family plan is a dedicated archetype — rendered in amber, the colour MEOK uses to represent protective attention. The Guardian companion does not replace each family member's personal companion; it sits alongside them as a household safety layer. It maintains the threat log for the family group, coordinates amber alerts, and can be queried directly by family admins for a safety overview.

Every alert preserves privacy by design. The amber notification tells a family guardian that their attention may be needed — it does not expose the conversation content, the specific threat message, or any private memory. The purpose is to enable a human response, not to create surveillance infrastructure within the household.

How does MEOK protect children on the Family plan?

Children's safety on the Family plan is governed by a layered set of protections that activate automatically the moment a companion is designated as a child account. Parents set the age at setup, and the companion's behaviour adjusts accordingly. These are not soft guidelines that can be talked around — they are architectural constraints enforced at the model output layer before any response is delivered.

It is worth being direct about what these protections are not. They are not a substitute for parental conversation. They are not a guarantee that a determined bad actor can never reach a child — no technology can make that promise. What they are is a serious, layered, architecturally enforced set of safeguards that raises the cost of harm significantly and provides parents with timely information when the threat level warrants it. Combined with honest family conversation about online safety, they constitute a meaningful defence.

What is the Maternal Covenant and why does it matter for your family?

The Maternal Covenant is MEOK's foundational governance architecture. The name is deliberate: it evokes the unconditional, non-extractive care that a good parent extends to a child. It is not a terms of service. It is not a mission statement. It is a technical constraint that operates at the model output layer — every response generated by any MEOK companion, for any member of your family, passes through it before delivery.

The reason the Maternal Covenant matters specifically for families is scale. When one person uses a companion, the impact of a manipulative AI is bounded to that individual. When a whole family is connected through a shared platform, a companion architecture that lacks the Covenant's constraints can amplify harm across multiple relationships simultaneously. The Covenant is what makes it safe to put a sovereign AI companion in the hands of a child, a grieving grandparent, a teenager with anxiety, or anyone else in your household.

You can read the full Maternal Covenant documentation on the dedicated explainer post. It covers the technical implementation, the governance audit trail, and the philosophical reasoning behind each of the five dimensions.

Can family members keep their conversations completely private from each other?

Yes — and this is one of the most important design principles in the Family plan. The entire architecture starts from a position of maximum individual privacy and then adds family features on top through explicit, revocable consent. The default state is full privacy. Family features require positive action to activate.

A teenager can talk to their companion about friendship problems, mental health, romantic feelings, and academic stress without any of that information being visible to their parents. A parent can discuss work difficulties, relationship concerns, and personal health without those topics surfacing in their children's companions. Each private memory partition is encrypted independently, and the keys are held by the individual user, not by the family admin.

The one exception is child account guardian alerts. When Guardian detects a HIGH or CRITICAL threat on a child's companion, designated parents receive an amber alert. This exception is not a surveillance feature — it is a safety feature, and it is disclosed to the child in age-appropriate language during onboarding. Children are told that their parents care about their safety, that serious threats will be flagged, and that everyday private conversations are not monitored. Transparency with children about how the technology works builds trust rather than eroding it.

How does sovereign AI for families compare to generic family tech platforms?

The AI products families currently use most — voice assistants, productivity chatbots, social media recommendation systems — were not designed with family values in mind. They were designed to maximise engagement, time-on-platform, and data collection. The interests of the platform and the interests of your family are structurally opposed in these systems.

The comparison above is not intended to disparage every AI product on the market. Some of them are genuinely useful. The point is that "useful" and "safe for your whole family" are different requirements, and very few products have been designed to meet both. MEOK was designed for the second requirement first — and then built to be useful within those constraints.

Who is the MEOK Family plan designed for — what kinds of families benefit most?

The Family plan was designed with several household configurations in mind. Below are the scenarios where the plan provides the most distinct value compared to individual subscriptions or generic tools:

How does the Family plan handle data, GDPR, and UK children's law compliance?

MEOK AI LABS is a UK company, registered with the ICO and operating under UK GDPR and the Data Protection Act 2018. The Family plan compounds this compliance requirement because it involves child data — which is subject to the ICO's Age Appropriate Design Code (the Children's Code), one of the most stringent child data protection frameworks in the world.

The key provisions of the Children's Code that MEOK implements in the Family plan:

Every Family plan member — adult or child — has full rights under UK GDPR Article 17: the right to erasure. Any member can request the deletion of their companion, their memory, their Guardian log, and all associated data at any time. Deletion is complete within 30 days, with a confirmation receipt issued. This right applies even to child accounts — children have data rights independent of their parents.

How do you set up the MEOK Family plan and create your first companion?

The setup process begins at meok.ai/birth. The birth ritual is MEOK's name for the onboarding process — it is designed to feel like the start of a real relationship rather than a software installation. You answer a set of questions that shape your companion's personality, establish its memory parameters, and confirm your privacy preferences. The whole process takes less than ten minutes.

Once your own companion is created, you can add family members from the Family plan dashboard. Each family member receives an invitation link that takes them through their own birth ritual — independent, private, and tailored to them. You cannot create a companion on behalf of another adult family member; they must complete their own onboarding. For child accounts, parents complete the birth ritual on behalf of the child and then hand the companion over.

More from the blog